We propose to obtain cross-sectional and longitudinal data on "HIV serosorting" among men who have sex with men (MSM) in San Francisco. We broadly define HIV serosorting as diverse strategies to reduce HIV acquisition or transmission by intentionally selecting sexual partners of the same serostatus or by modifying sexual practices depending on the partner's serostatus. The current phase of the HIV epidemic in San Francisco is complex with rising levels of overall unprotected anal sex (UAS) and sexually transmitted diseases (STD), yet stable HIV incidence. We hypothesize that HIV serosorting may explain these apparently contradictory findings;that is, increasing UAS with partners of the same HIV serostatus is leading to increases in STD incidence but not HIV incidence. While some evidence suggests serosorting may be increasing among MSM in San Francisco, many questions remain: How prevalent is serosorting as a deliberately adopted HIV prevention strategy in the MSM community? How do MSM identify partners of the same serostatus? What factors contribute to the success or failure of adhering to serosorting strategies? We need detailed and prospectively collected data in order to answer these questions and to assess the causal relationship between reported serosorting and actual risk for HIV. Accurate information is essential for dispelling misperceptions of serosorting if it is not causally associated with reducing serodiscordant UAS, or for increasing serosorting success if it is, or for framing serosorting in the context of risk reduction (but not elimination) if its in role in reducing risk is mixed. We propose to recruit a community-based cohort of HIV- negative and HIV-positive MSM using a probability-based time-location sampling method with longitudinal follow-up over the internet.